SoC Implementation Tool Yields Speedier Frequency Results

A revise of AmmoCore's Fabrix system-on-a-chip (SoC) implementation suite improves quality of results, promises reduced die sizes, and aims to shorten design cycles. The suite combines floorplanning and silicon virtual prototyping with physical implementation.

Fabrix is built around the company's SBlock architecture, which packs from 300 to 3000 standard cells into SBlock modules. A new SBlock placement algorithm shortens wire lengths for improved timing and less congestion. Also, automatic boundary-scan-cell placement improves I/O timing. Clock speedups of 10% have been seen compared to customers' previous methodologies, according to AmmoCore.

The SBlock architecture's fluidity provides more efficient area usage. A novel SBlock sizing algorithm further reduces area by compacting SBlocks to an optimal size without affecting routability. Die-size improvements of as much as 15% have been obtained.

A new seed and hard-macro placer enhances productivity during floorplanning. It automatically clusters and places macros in close proximity to their associated logic with the correct macro orientation. The seed placer performs a near-manual placement.

Fabrix costs $1.2 million for a one-year term license for Solaris and Linux platforms.